Restricted Environments and Incentive Compatibility in Interdependent Values Models

Abstract

We study the possibility of designing satisfactory ex post incentive compatible single valued direct mechanisms in interdependent values environments, characterized by the set of agents' type profiles and by their induced preference profiles. For environments that we call knit and strict, only constant mechanisms can be ex post (or interim) incentive compatible. For those called partially knit, ex post incentive compatibility extends to groups, and strategy- proofness implies strong group strategy-proofness in private values environments. The results extend to mechanisms operating on non-strict domains under an additional requirement of respectfulness. We discuss voting, assignment and auctions environments where our theorems apply.